Professional Documents
Culture Documents
1
p
.
In general, H(f) is a complex quantity called Fourier transform of
h(t). H(f) can also be expressed as
R
2
f I
2
f
q
F f tan
1
I f
R f
When the amplitude spectrum of h(t) is studied, it is
possible to identify the principal harmonic components of the
digital record, because they always appear with major ampli-
tude [16]. Frequencies corresponding to principal harmonic
components are the frequencies of the tide waves present in the
register. These waves are identied visually and their period T
(T = 1/f) can be read directly from the amplitude spectrum.
Thus, these harmonic components are easily identied in
the amplitude spectrum. Obviously, spectral analysis can be
used to identify and quantify any harmonic component in a
digital register. For this reason, the spectral analysis is
considered today a standard tool of analysis in many scientic
elds.
Table 1
FT Analysis of EMAs in Mexico.
ID Name station State Period obtained (Hz)
Cal Calvillo Aguascalientes 1.157e005
PELZ Presa Emilio Lo pez Zamora Baja California Norte 1.157e005
CSL Cabo San Lucas Baja California Sur 1.157e005
CDC Ciudad del Carmen Campeche 1.157e005
Pal Palenque Chiapas 1.157e005
Bas Basaseachi Chihuahua 1.157e005
NRo Nueva Rosita Coahuila 1.157e005
ENCB Escuela Nacional de Ciencias Biolo gicas Distrito Federal 1.157e005
Ame Agustn Melgar Durango 1.157e005
PAll Presa Allende Guanajuato 1.157e005
Cal Ciudad Altamirano Guerrero 1.157e005
Pac Pachuca Hidalgo 1.157e005
Tiz Tizapan Jalisco 1.157e005
PMa Presa Madn Me xico 1.157e005
Ang Angamacutiro Michoaca n 1.157e005
IMTA Instituto Mexicano de Tecnologa del Agua Morelos 1.157e005
Aca Acaponeta Nayarit 1.157e005
PEC Presa El Cuchillo Nuevo Leo n 1.157e005
PAn Puerto A
Fig. 2. Example of the data (Presa Emilio Lo pez Zamora in Baja California Norte state from 1st-01-2007 to 31st-12-2007).
Q. Hernandez-Escobedo et al. / Renewable and Sustainable Energy Reviews 15 (2011) 721728 723
2.2.2. Characterization of the average day
To characterize the average day for each station in the years
2007 and 2008, we have developed an Excel software application
that calculates the average of each 10 min interval for the 24 h to
nally obtain the average day of the year. This information is
important for the statistical comparison of areas.
3. Theory/calculation
The study of the frequency of wind requires the capture and
management of a large amount of data, a sampling period of a few
minutes and a period of observation of the order of years to
appreciate daily and seasonal periodicity.
The direct application of FFT to the entire data block captured
fromthe wind will not provide very valuable information. It will be
necessary to develop a specic process of treatment of these
signals that denote the characteristic parameters that can be
expected, such as if the existence of periodicity and what type of
variations occur along a period of a year.
A sampling period Tmof 10 min in the capture of data allows to
collect the most important information of the variable, but
generates sets of the order of 50 KS per year.
The direct application of the FFT to the set of samples would
provide spectral information that would not reect seasonal
changes, so we have developed the following specic technique of
analysis.
The set of samples collected of the variable is fragmented by
means of windows of time of appropriate type. A rectangular
window introduces distortions due to convolution with TmSinc(f
Tm) in the domain of frequency. Distortion increases the
importance of a less number of samples of the window [19].
Window Hamming is the most suitable for this application. The
temporal width of the window is set to: Tmk, obtaining a set of
unidimensional vectors of size k of the variable. Then we apply the
FFT to each vector, which produces two vectors of size k,
corresponding to the module phase respectively. By means of a
process of decimation that eliminates the elements of the FFT
image we get two vectors of size K/2: espec[t = tmk0],
espec[t = tmk1] as shown in Fig. 3.